On 2010-05-08, Uwe Klein <[email protected]> wrote: > Rob wrote: >> Defending silly algorithms in code with limitations in old hardware no >> longer in use in running systems seems a bit silly to me. > > > Any abstraction _must_ cover all variations that are sensible or > expectable.
While one might regard it as desireable that they cover the possible systems, that does NOT mean that the program has to run just the lowest common denominator. It should test to see if the system is one of the brain damaged ones, and then run a special case, not assume that because one in a million is brain damaged, all must be treated as if they are. > An abstraction that only covers the smallest common set of features > and that attaches to the available set of most extended limitations > is just broken. > > An abstraction should have a builtin mechanism for extensions > ( and be "innoculated" against possibly resulting breakage.). > > uwe _______________________________________________ questions mailing list [email protected] http://lists.ntp.org/listinfo/questions
